gs_horizon

[lucius81]

Hi Federico, I would like to know if the gs_horizon is meant to compare enrichment analyses resulting from independent DEG analyses. In my case I am performing three different pairwise comparisons between clones of the same cultivar grown under the same conditions, therefore most enriched processes are shared among the three comparisons.

For each pairwise comparison i have performed independent deseq2 analysis. Afterwards I have performed a geneProfiler analysis, followed by the GeneTonic shaker and aggregate scores functions.

My question is: Is it possible to summarize these three pairwise comparisons into a gs_horizon graph?

I was not able to adapt the instructions shown here: https://federicomarini.github.io/GeneTonic/reference/gs_horizon.html to my datasets..

Thanks in advanced,
Luciano

[federicomarini]

Hi Luciano,
The gs_horizon should indeed be the way to go.
Running the example lines

gs_horizon(res_enrich,
           compared_res_enrich_list = compa_list,
           n_gs = 50,
           sort_by = "clustered"
)

... which would be the unsquished version of what you (don't really) see in the link of the rendered documentation.

You can also play with the sort_by parameter to get a different ordering of the terms

You can focus on the compa_list object, that is pretty much the form you should bring the other enrichment results into.

[lucius81]

Hi Federico, thanks for your answer.

My problem is that after defining my data frames and running the gs_horizon function, i keep getting the same error message (both res_enrich and res_enrich2 have gone through the shaker and aggscores functions. Therefore, they both are 11 columns data frames).

res_enrich <- res_enrich_all_greater1_M [1:18, ]
res_enrich2 <- res_enrich_all_greater1_P [1:18, ]
compa_list <- list(scenario2 = res_enrich2)

gs_horizon(res_enrich,
compared_res_enrich_list = compa_list,
n_gs = 10,
p_value_column = "gs_pvalue",
color_by = "z_score",
ref_name = "ref_scenario",
sort_by = "first_set"
)
Warning messages:
1: Removed 10 row(s) containing missing values (geom_path).
2: Removed 10 rows containing missing values (geom_point).

This is my disappointing output image:

Off course I am certain that there are shared processes between the two compared sets.. so i cannot tell where i am messing-up..

thanks again,
Luciano

[federicomarini]

Do you have a snapshot of the necessary objects (ideally the .Rdata workspace?) for me to have a (confidential) look at?
If so, we can continue via mail (see the one listed as maintainer in the package DESCRIPTION)
Federico